The production of three-dimensional photographs, both cine and still, is well known in the art and usually involves the use of two cameras which record the same scene from slightly different positions. Typically, the nominal effective lateral displacement between the respective optical axes of the cameras, known as the interaxial spacing, is approximately equal to the average spacing between the eyes of a viewer (i.e. about 2.5 inches). This spacing would produce the same 3-D effect as that perceived by a viewer who viewed the scene from the camera location. A greater or lesser interocular spacing is sometimes desirable to produce greater or lesser depth. Hence, conventional 3-D camera systems commonly include some means for adjusting the position of one camera relative to the other to vary the interaxial spacing over a continuous range from zero to about 4 or 5 inches.
FIGS. 4a-c are stereoscopic-pair drawings to show the change on stereoscopic depth of a scene as a result of increasing the interaxial spacing (using a 3-D camera rig of this invention as the subject of the drawings). These stereo drawings are intended to be free viewed, with the left images on the left, right images on the right (not cross eyed). When viewed in 3D, FIG. 4a shows the stereoscopic depth as if photographed with a 0.5-inch interaxial spacing. FIG. 4b shows the same subject as if photographed with a 2.5-inch spacing (equal to the distance between our eyes, and therefore more normal depth). FIG. 4c shows exaggerated stereo depth as if photographed at 4.5-inch interaxial spacing. Interaxial adjustment is one of two primary controls that a stereographer needs to shoot a 3-D movie (the other adjustment being the convergence distance, see FIG. 5).
Aside from interaxial spacing adjustment, the off-screen position of images can also be controlled by varying the convergence distance at which the optical axes of the respective 3-D camera lenses intersect. When 3-D images are projected for viewing, such convergence distance determines the distance at which objects within the scene appear to be located front-to-back relative to the projection screen. When the convergence angle is such that the camera axes intersect at, say, thirty feet in front of the cameras, objects in the foreground and background will appear to a viewer to be positioned in front of and behind the screen, respectively. Special well known 3-D effects can be achieved in 3-D motion picture photography by varying the convergence distance during filming, and conventional 3-D camera systems commonly include means for adjusting the convergence distance of the lens axes of the two cameras to vary such distance over a continuous range between infinity and about four feet measured from the camera.
FIGS. 5a-c are stereoscopic-pair drawings with a fixed amount of stereo depth, or amount of 3D, using a 2.5-inch interaxial spacing, however showing the effect of changing the convergence distance. For reference, any object photographed at the convergence distance (the distance where the two camera lens axes intersect in space) will be seen by the audience on the surface of the theater screen. Objects farther than the convergence distance will appear farther, behind the screen. Objects closer than the convergence will appear closer than the screen, floating out toward the audience. It is important for 3-D camera rigs to provide the ability to change convergence distance to allow the camera operator to push and pull the image in or out of the screen. FIG. 5a is drawn as if the right and left cameras converged on the front corner of the object. This places the image at the screen plane (the plane of the paper) and behind. FIG. 5b is drawn with “cameras” converged at an intermediate depth in the scene, so that the front part of the image comes off the paper, and the back part recedes into the paper. FIG. 5c is drawn with the convergence distance at the back of the scene, so that the entire image protrudes off the paper.
There are three basic approaches to stereoscopic or 3-D photography.
The most basic approach for stereoscopic or 3-D photography has been done with single-lens adapters. The primary advantage of this approach is that it is simple because it uses a single camera, reducing the cost. The disadvantages of this approach are that the interaxial spacing is fixed, restricted by the size of the lens, and frequently the convergence distance is fixed. The left and right images are squeezed onto a single frame of film, compromising the image quality.
Another way stereoscopic or 3-D photography has been done is with side-by-side cameras as illustrated in FIG. 1 in which a 1-foot long ruler is shown for reference. The advantages of this approach are no light loss to the cameras, compared to beamsplitter rigs, and that such rigs are simple and inexpensive. The disadvantage of this approach is that the minimum interaxial spacing can be no less than the width of the camera. For example, Panavision 35 mm film cameras are approximately 10.5 inches wide, although some new video cameras are as narrow as 3 inches. However, even a 3-inch interaxial spacing is too wide for most shots. Therefore, side-by-side rigs are not practical for most 3-D work.
A third way stereoscopic or 3-D photography has been done is with beamsplitter rigs as illustrated in FIGS. 2a-c. FIG. 2a shows the laterally adjustable camera DC parallel to (no convergence) and beside the optically equivalent position of the stationary camera RC. Camera DC is spaced laterally at an interaxial spacing IAX from the optically equivalent position of the stationary camera RC. The camera lens axes are parallel and are considered to converge at infinity. FIG. 2b shows the direct camera DC at the same lateral position (interaxial spacing) as in FIG. 2a; however, toed in at a slight angle so that the lens center line converges with stationary camera RC at a distance, less than infinity but outside of the left side of the drawing. FIG. 2c shows the direct camera DC at the same angle as in FIG. 2b; however at a reduced interaxial spacing. Notice that the convergence distance has unintentionally and undesirably been reduced as well. This conventional 3-D rig design lacks the independence between adjustments of interaxial and convergence.
The advantages of using a beamsplitter rig are that the laterally-adjustable camera DC can be positioned in the optically-equivalent position of the fixed camera RC providing a minimum interaxial spacing of as little as 0.0 inches. Disadvantages are the 50% light loss to both cameras from the beamsplitter, the large size of the beamsplitter required due to the horizontal field of view of the laterally adjustable camera DC in its maximum interaxial position, and the difficulty of rigidly mounting the thin glass beamsplitter supported on its bottom edge. The large size beamsplitter is required because the camera lens sees the scene as a wider-than-high rectangle in front of the camera. Light from the scene converges into the lens and passes through the lens' entrance pupil, well inside the lens, where the light rays are at their smallest diameter.
FIGS. 6-9 show a variety of conventional camera lenses, drawn at the same scale, used in the motion picture and television industries. In all cases, note the position of the entrance pupils EP being well inside the lens barrels at a distance of DEP, and the clear area CA at the front of the lenses, required if a mirror or beamsplitter is used in a 3-D rig application. Because traditional lenses used in the motion picture industry are physically large, and have the entrance pupil inside the lens barrel, and because the beamsplitter must be located out in front of the lens by several inches, a sizeable portion of the beamsplitter is required to cover the image as it passes through, or is reflected off of, the beamsplitter. The physical scale of the cameras, lenses and beamsplitter limits how small a beamsplitter-type 3-D camera rig can be.
Because traditional lenses used in the motion picture industry are physically large, and have the entrance pupil inside the lens barrel, and because the beamsplitter must be located out in front of the lens by several inches, a sizeable portion of the beamsplitter is required to cover the image as it passes through, or is reflected off of, the beamsplitter. The physical scale of the cameras, lenses and beamsplitter limits how small a beamsplitter-type 3-D camera rig can be.
FIGS. 3a-c show a state-of-the art dual-camera beamsplitter 3-D rig fitted with video cameras. The physical size of the cameras and lenses, and lens field of view and position of entrance pupil deep in the lens causes the rig to be rather large. A 3-foot ruler is included for reference. The advantages of this approach are the narrow rig with wide-angle lenses, rigid structure, that the cameras interaxial spacing can be adjusted to as little as 0.0 inches. The disadvantages are the physical size of the rig, the weight, and the light loss due to the 50/50 beamsplitting mirror.
The ease of use of rigs used in stereoscopic or 3-D photography today is limited by physical size limitations or compromises in versatility and adaptability that are not generally acceptable. Accordingly, there has been a long felt need in 3-D photography for new and improved rigs that are smaller, lighter and easier to use while still permitting interaxial spacing and/or convergence distance adjustments desired by filmmakers.